基于遗传算法的机器人自定位

为了解决中型组比赛环境下的足球机器人自定位、绑架和跟踪问题,提出一种基于遗传算法的机器人连续自定位方法.首先,从全景图像中提取出场上黑线与机器人辐射线的交点;然后,把该交点与环境地图对应黑线间距离之和的最小值做为适应度函数;最后,以机器人中心坐标做为个体,进行选择、交叉和变异操作,收敛到机器人的自定位坐标.另外,针对绑架和连续定位,在遗传算法全局自定位基础上利用梯度优化算法局部修正主位姿,提高了自定位精度和算法的鲁棒性.仿真结果和场地环境机器人自定位试验说明了该定位方法的有效性.     

A practical self-localization scheme for mobile robots using sonar sensors

A practical self-localization scheme for mobile robots is proposed and implemented by utilizing sonar sensors. Specifically, the localization problem is solved by employing Monte Carlo method with a new mechanism proposed to calculate the samples' weights; the convergence and veracity of the sample set are guaranteed by the designed resampling and scattering process. The proposed self-localization algorithm is fully implemented on a specific mobile robot system, and experimental results illustrate that it provides an efficient solution for the kidnapped problem.